Automatic Topic Learning for Personalized Re-Ordering of Web Search Results

Abstract

The fundamental idea behind personalization is to first learn something about the users of a system, and then use this information to support their future activities. When effective algorithms can be developed to learn user preferences, and when the methods for supporting future actions are achievable, personalization can be very effective. However, personalization is difficult in domains where tracking users, learning their preferences, and affecting their future actions is not obvious. In this paper, we introduce a novel method for providing personalized re-ordering of Web search results, based on allowing the searcher to maintain distinct search topics. Search results viewed during the search process are monitored, allowing the system to automatically learn about the users' current interests. The results of an evaluation study show improvements in the precision of the top 10 and 20 documents in the personalized search results after selecting as few as two relevant documents.

A Luhn-Inspired Vector Re-weighting Approach for Improving Personalized Web Search

“…The noise in the profile may cause some of search results which are irrelevant to the user's current search interests to be promoted in the search results list, and reduce the effectiveness of the personalization. miSearch [9] addresses this problem by allowing searchers to create distinct search topic profiles. This system automatically learns the searchers' interests based on search results that were viewed in the course of previous searches on the topic.…”

Personalised Web Search using Browsing History and Domain Knowledge

“…Information stored in user profile can be used to disambiguate or to infer user's query context. Studies in personalized search include [6] which provided the searcher with different search topics and monitored clicked search results so as to learn user's current interests and re-order web search result accordingly. Another approach in [5] models user interests as a vector of weighted terms from visited URLs, and apply a snippet scoring method to re rank search results.…”